The Vinculin Binding Sites of Talin and α-Actinin Are Sufficient to Activate Vinculin*

Vinculin regulates both cell-cell and cell-matrix junctions and anchors adhesion complexes to the actin cytoskeleton through its interactions with the vinculin binding sites of α-actinin or talin. Activation of vinculin requires a severing of the intramolecular interactions between its N- and C-terminal domains, which is necessary for vinculin to bind to F-actin; yet how this occurs in cells is not resolved. We tested the hypothesis that talin and α-actinin activate vinculin through their vinculin binding sites. Indeed, we show that these vinculin binding sites have a high affinity for full-length vinculin, are sufficient to sever the head-tail interactions of vinculin, and they induce conformational changes that allow vinculin to bind to F-actin. Finally, microinjection of these vinculin binding sites specifically targets vinculin in cells, disrupting its interactions with talin and α-actinin and disassembling focal adhesions. In their native (inactive) states the vinculin binding sites of talin and α-actinin are buried within helical bundles present in their central rod domains. Collectively, these results support a model where the engagement of adhesion receptors first activates talin or α-actinin, by provoking structural changes that allow their vinculin binding sites to swing out, which are then sufficient to bind to and activate vinculin.

Adhesion complexes form on the cell membrane when cells come in contact with each other or with components of the extracellular matrix and trigger elaborate signaling networks that direct dynamic and rapid rearrangements of the actin cytoskeleton (1,2). Vinculin is a highly conserved and critical regulator of both cell-cell and cell-matrix junctions, as it anchors these junctions to the actin cytoskeleton by binding to talin and ␣-actinin, which directly interact with integrin and/or cadherin transmembrane receptors in focal adhesions and adherens junctions, respectively (3,4). Overall, vinculin appears to stabilize these junctions, as vinculin overexpression augments the formation of adhesion complexes and prevents cell migration (5), whereas vinculin loss leads to marked defects in adhesion with the extracellular matrix, cell spreading, and concomitant increases in the rates of (chaotic) cell migration (6,7). Furthermore, vinculin appears to play an important role in cancer, where it functions as a tumor suppressor that inhibits cell invasion and metastasis (8), and in other pathophysiological scenarios, including wound healing, ischemia, and apoptosis (9 -11).
Acidic phospholipids such as phosphatidylinositol-4,5-bisphophate (PIP 2 ) 2 have been thought to serve as the triggers that activate vinculin (21,25), and indeed PIP 2 can bind to and alter the conformation of the Vt domain (26). However, PIP 2 interactions with vinculin are disrupted at physiological salt concentrations, and its binding to vinculin requires high levels of PIP 2 in lipid micelles (27,28). Furthermore, PIP 2 competes with F-actin for binding to vinculin (29), and the PIP 2 binding site is occluded in the full-length structure of inactive vinculin (15). Finally, vinculin mutants selectively defective in PIP 2 binding still recruit to sites of focal adhesions (28). Thus, although PIP 2 may play some role in regulating vinculin, other triggers likely activate vinculin.
Mounting evidence suggests that talin and ␣-actinin may directly activate vinculin. During outside-in integrin signaling, the first event detected is PIP 2 -mediated changes in the conformation of the head domain of talin (30), which then rapidly associates with the cytoplasmic tails of ␤-integrin receptors (31)(32)(33). Talin then interacts with vinculin through several high-affinity vinculin binding sites (VBS) present in its central rod domain (16, 34 -36), which could allow talin to bind to multiple vinculin molecules and amplify outside-in integrin signaling (35). These interactions are likely essential, as targeted deletion of talin abolishes the formation of focal adhesions (37). Similarly, ␣-actinin plays an important role in the maturation of adhesion complexes (38,39) and binds to vinculin through a single high affinity VBS (␣VBS) present in the R4 spectrin repeat at the end of its rod domain (19,20,40). Notably, the ␣VBS and VBSs of talin efficiently disrupt the Vt domain from pre-existing Vh1⅐Vt complexes (24,35,41). Moreover, the crystal structure of inactive human vinculin (12), those of Vh1⅐talin-VBS complexes (24,35,36,41), and that of the Vh1⅐␣VBS complex (40), have demonstrated that the binding site for talin or ␣-actinin in vinculin is not masked by the head-tail interactions of vinculin; rather, it is readily accessible to the VBSs of talin or ␣-actinin (12). Finally, the structures of talin-VBS-and ␣VBS-bound to Vh1 have demonstrated that these amphipathic ␣-helices disrupt the Vh1-Vt interaction from a distance, by provoking conformational changes in the N-terminal helical bundle of vinculin, by a process coined helical bundle conversion (24,35,40).
The structures of native, inactive talin and ␣-actinin have revealed that their VBSs are normally buried within helical bundles that comprise their rod domains (40 -43). Thus, in their resting state, the rod domains of talin or ␣-actinin have a rather low affinity for vinculin (15,20), and this has suggested a combinatorial model for vinculin activation, where simultaneous binding of two of more ligands are required to provide the free energy necessary to break the head-tail interactions of vinculin (15,18). However, this model only considers talin or ␣-actinin in their inactive states and the structure of the helical bundle domains of talin or ␣-actinin may also be dynamic as, for example, atomic force microscopy has shown that helical bundle domains that comprise spectrin repeats (also found in ␣-actinin) can form stable, unfolded intermediates when exposed to mechanical stress (44). Thus, when exposed to tension forces from within or from outside the cell, such as occurs during the formation of adhesion complexes (45)(46)(47)(48), the VBSs present in the helical bundle domains of talin and ␣-actinin might become exposed to bind to and activate vinculin, without the need for co-stimulatory signals.
Because talin and ␣-actinin do associate with vinculin at adhesion sites in cells (49 -51), and ␣VBS has a comparably high affinity for fulllength vinculin versus the isolated Vh1 domain (K d of ϳ2 nM; Ref. 40), we predicted that physiological levels of the VBSs of talin and ␣-actinin might be sufficient to activate vinculin. Indeed, here we present biochemical and biological tests of this hypothesis, which support the notion that the VBSs of talin and ␣-actinin function as physiological triggers of vinculin activation in adhesion complexes.
Surface Plasmon Resonance Assays-Binding studies were performed by surface plasmon resonance (SPR) using a Biacore 2000 biosensor equipped with a carboxymethyldextran-coated gold surface (CM-5) sensor. The carboxymethyl groups on the chip were activated with EDC and N-hydroxysuccinimide to form the N-hydroxysuccinimide ester of carboxymethyldextran. Vinculin protein was attached to this activated surface by reaction of the carboxyl groups of dextran with the primary amines of vinculin to form an amide linkage. Any remaining reactive sites on the surface were blocked by a reaction with ethanolamine. Reference cells were prepared similarly except that no vinculin protein was added. The interaction of vinculin with the VBSs of talin was analyzed at 25°C in HPBS-P buffer (10 mM HEPES (pH 7.4), 150 mM NaCl, 0.1 mg/ml bovine serum albumin, 0.005% P-20). Talin-VBS1, -VBS2, and -VBS3 peptides were solubilized in water and then diluted in HPBS-P buffer. Binding was measured by flowing the individual talin-VBS peptides at a flow rate of 15 l/min through the reference and vinculincontaining flow cells in sequence. Blanks were also run consisting of only buffer. Association was measured for 10 min, and dissociation was measured for 15 min. Data reported are the difference in SPR signals between the flow cells containing vinculin and the reference cells. Any contribution to the signal was removed by subtraction of the blank (buffer) injection from the reference-subtracted signals. Responses at equilibrium were then plotted as a function of the peptide concentration to calculate the affinities of the interactions of the VBSs of talin for full-length vinculin using Scrubber software.
Vinculin-VBS Binding Assays-Recombinant protein comprising the entire vinculin head domain (VH, residues 1-840), which includes residues in the Vt2 domain that have been suggested to contribute to the head-tail interactions of vinculin (18), was incubated with Vt (residues 879 -1,066) for 20 min. To ensure that all VH was driven into a complex with Vt, the VH⅐Vt complex was formed using a 1:2 molar ratio of VH:Vt. An analysis on native polyacrylamide gels confirmed the rapid and complete formation of the VH⅐Vt complex under these conditions, which was distinguishable from free VH protein. ␣VBS or talin-VBS3 peptides were then added to the VH⅐Vt complex at the indicated molar ratios (from 1:1 to 20:1, VBS:vinculin) and allowed to incubate for 20 min. Complexes formed were then resolved on 8 -25% gradient native polyacrylamide gels. The identity of components of the complexes in native gels was confirmed by cutting out the bands and analyzing on SDS-polyacrylamide gels and by immunoblotting with antibody specific for the histidine tag on vinculin (data not shown).
Vinculin-Actin Binding Assays-Vinculin was first incubated for 20 min at ambient temperature with talin-VBS3 or ␣VBS peptide (in 20 mM Tris-HCl (pH 8.0), 0.2 mM CaCl 2 , 0.2 mM ATP, 2 mM MgCl 2, and 100 mM KCl) at the indicated molar ratios (from 1:1 up to 20:1, VBS: vinculin) in a final volume of 580.8 l. F-actin was polymerized as described (29) in a 100-l total volume of the same solution, supplemented with 5% (w/v) sucrose and 1% (w/v) dextran. 19.2 l of polymerized F-actin (final concentration of 7.5 M) was then added. These complexes were then incubated for 1 h (at ambient temperature, a final total reaction volume of 600 l), and the samples were then sedimented at 100,000 ϫ g at 25°C in a Beckman ultracentrifuge for 15 min. Equal volumes of reaction pellets (P) and supernatants (S) were resolved on 7.5% SDS-polyacrylamide gels and the gels were stained with Coomassie Blue.
NMR Spectroscopy-All NMR experiments were carried out at 25°C on a Bruker DRX800 spectrometer equipped with a 5-mm HCN/z probe. Each 1 H, 15 N TROSY experiment was recorded for 24 h. Product ion (MS2) spectra were subjected to search using the SEQUEST program of Eng and Yates (ThermoQuest). 98% 1 H, 15 N-labeled, octahistidine-tagged, full-length human vinculin (residues 1-1,066) protein was expressed in Escherichia coli strain BL-21 in a culture of M9 minimal medium (D 2 O) supplemented with 1 g of 15 N-labeled ammonium chloride and 4 g/liter glucose. 2 H, 15 N-labeled vinculin was purified as previously described (12) and was dialyzed against 20 mM potassium phosphate buffer (pH 7.6) and was concentrated to 0.2 mM. NMR spectroscopy of the vinculin⅐␣VBS and the vinculin⅐talin-VBS3 complexes were obtained following mixing of a 1.4:1 molar ratio of these VBSs to vinculin.
Cell Culture-Swiss-3T3 cells were obtained from the American Type Culture Collection (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, penicillin, streptomycin, and L-glutamine in 9% CO 2 . Vinculin Ϫ/Ϫ and vinculin ϩ/Ϫ cells were the generous gift of Eileen Adamson (The Burnham Institute, La Jolla, CA) and were cultured as previously described (7).
Microinjection, Immunofluorescence, and Video Microscopy-For microinjection, cells were plated onto poly(D-lysine)-coated 35-mm glass bottom dishes (MatTek Corp., Ashland, MA) the day before injec-tion in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and were incubated at 37°C, 9% CO 2 . Because of their rather poor adhesive properties (7), vinculin Ϫ/Ϫ cells were plated onto poly(D-lysine)-coated 35-mm glass bottom dishes that were also coated with gelatin. For direct comparison vinculin ϩ/Ϫ cells were similarly plated on glass dishes coated with gelatin. All microinjections were performed using 1ϫ injection buffer (50 mM HEPES, 100 mM KCl, 40 mM NaHPO 4 (pH 7.2)). Cells were then microinjected with the indicated VBS peptides, injecting ϳ1,000 or ϳ10,000 molecules/cell using an Eppendorf microinjection system (Brinkmann, NY), mounted on an Axiovert 135 TV microscope (Carl Zeiss), with a 32ϫ Achrostigmat phase-contrast objective (NA 0.4). This microinjection system was calibrated to deliver, on average, 1.2 Ϯ 0.3 picoliters/pulse. Using the given peptide concentrations, we calculated the approximate number of molecules delivered per injection. The microscope heated stage was kept at 37°C, and the injection parameters were P i : 20 -40 hPa, T i : 0.5 s, and P c : 20 hPa. As a control, cells were injected with 1ϫ injection buffer alone. An image of each cell about to be injected was taken prior to its injection with the needle pointing at the cell. To indeed confirm correct delivery of the peptides into the cells, they were co-injected with Alexa Fluor 568 (55 ng/ml, from Molecular Probes). After all the cells were injected, images were taken every 5 or 10 s, for the next 10 -15 min. The acquisition time was stamped on each image, and all the images were combined in an AVI animated file.
To visualize the actin cytoskeleton and the localization of cytoskeletal proteins, Swiss-3T3 cells were plated and microinjected on Permanox lab-tek chamber slides (Nalge Nunc Corp., Napeville, IL). Immediately after injection, the cells were fixed for 10 min in 4% paraformaldehyde 1ϫ phosphate-buffered saline. The cells were then permeabilized for 10 min with 0.1% Triton X-100 in phosphate-buffered saline and were then blocked with 3% milk in phosphate-buffered saline prior to applying the primary antibodies. Actin filaments were identified by staining the cells with TRITC-coupled phalloidin (Sigma). Focal adhesion complexes were identified by staining cells with antibodies that detect mouse vinculin, talin, and ␣-actinin (all from Sigma).

RESULTS
The VBSs of Talin Have High Affinity for Full-length Vinculin-The binding affinity of ␣VBS for full-length vinculin is similar to that for the Vh1 domain alone (K d of both are ϳ2 nM; Ref. 40). The affinity of talin-VBS3 for the Vh1 domain (residues 1-258) was in the same range (K d of 3 nM) and is greater than that of talin-VBS1 (K d of ϳ15 nM) or talin-VBS2 (K d of 32 nM) for Vh1 (35). However, additional interdomain contacts of Vt with Vt2 have been shown to compromise the binding of the entire rod domain of talin (residues 397-2,541; Ref. 18). To define the affinity of the VBSs of talin for full-length vinculin, we performed SPR binding assays for talin-VBS1, -VBS2, and -VBS3 to full-length human vinculin protein immobilized on a Biacore chip. Similar to its lower affinity interaction for the isolated Vh1 domain (35), talin-VBS2 had a relatively low affinity for full-length vinculin (Fig. 1, C and D). However, talin-VBS1 and talin-VBS3 had affinities (K d of 77 and 74 nM, respectively; Fig. 1, A, B, E, and F) that were comparable to that of the intramolecular interactions reported for the head and tail domains of vinculin (K d ϳ50 -90 nM; Refs. 17,18). Because the affinities of the VBSs  A, VH protein (residues 1-840, lane 1) was incubated with Vt (residues 879 -1,066) protein (at a 1:2 molar ratio, to assure complete association of VH with Vt); the resulting VH⅐Vt complex detected in native gels (lane 2) was confirmed to contain the ϳ90 and 30-kDa tail domains by SDS-PAGE analysis (data not shown) and was distinguishable from free VH (lane 1). Note that free Vt is not detectable on native gels because of its high pI. ␣VBS (A and B) or talin-VBS3 (C and D) peptides were then titrated into the VH⅐Vt complex and allowed to incubate for 20 min. Even relatively low molar ratios of these VBSs were effective at displacing Vt to form the indicated VH⅐VBS complexes, which were distinct from those of the VH⅐Vt complex or free VH on native gels. Again, the identity of the VH⅐VBS complexes were confirmed by SDS-PAGE analyses (data not shown). B and D, quantitation of the dissociation of the VH⅐Vt complex (solid lines) and the formation of the VH⅐␣VBS (B) and VH⅐talin-VBS3 (D) complexes (dotted lines) are also shown.
of talin for full-length vinculin were lower than that for the isolated Vh1 domain (24) additional intramolecular interactions within the fulllength molecule do impair talin-VBS binding to the Vh1 domain. However, their affinities for full-length vinculin are still quite high, and in the physiological range expected to effectively compete with the intramolecular interactions of the head and tail domains of vinculin.
The VBSs of Talin and ␣-Actinin Displace the Head-Tail Interactions of Vinculin-All of the VBSs of talin and ␣VBS bind to the Vh1 domain of vinculin in a mutually exclusive fashion, by inserting between helices ␣1 and ␣2 of the N-terminal helical bundle of the Vh1 domain of vin-culin, and all are capable of displacing Vt from pre-existing Vh1⅐Vt complexes (24,35). The high affinity of the VBSs of talin and ␣-actinin for full-length vinculin in SPR binding assays ( Fig. 1 and Ref. 40) suggested that the same would hold true in solution. However, others have shown that the interactions of Vt with both the Vh1 and Vt2 domains compromises binding of the entire talin rod to full-length vinculin in solution (18). To test whether talin-VBS3 and ␣VBS were sufficient to displace Vt from complexes formed between the entire head domain of vinculin (VH, residues 1-840) with Vt, recombinant VH and Vt proteins were incubated together (at a 1:2 molar ratio of VH:Vt); native gel analyses established the rapid and complete formation of the VH⅐Vt complex, which was distinguishable from free VH on these gels (Fig. 2,  A, lane 2 versus lane 1, and C, lane 1). To then test whether the VBSs of ␣-actinin and talin were sufficient to displace Vt from this complex, we titrated these VBS peptides into the head-tail complex. Even relatively low molar ratios of ␣VBS or talin-VBS3 were sufficient to displace Vt and to form novel VH⅐VBS complexes (Fig. 2, A-D). Similar findings were evident when testing the ability of talin-VBS1 to disrupt Vt from the VH⅐Vt complex (data not shown). Further, consistent with its higher affinity by SPR assays (40), ␣VBS was more effective at displacing Vt from the VH⅐Vt complex than talin-VBS3 or talin-VBS1 (Fig. 2, and data not shown). Therefore, in solution, the VBSs of ␣-actinin or talin are sufficient to disrupt the intramolecular head-tail interactions of vinculin, the initiating event of vinculin activation.
The VBSs of Talin and ␣-Actinin Alter the Conformation of Fulllength Vinculin-The binding of the VBSs of talin or ␣-actinin to fulllength vinculin alters its sensitivity to proteases, suggesting that these VBSs open up regions that are buried in inactive vinculin (40). To directly assess the effects of the VBSs of talin and ␣-actinin on the overall conformation of vinculin, we performed 1 H, 15 N TROSY NMR spectroscopy of full-length vinculin (Fig. 3B) and of vinculin when bound by talin-VBS3 (Fig. 3A) or ␣VBS (Fig. 3C), at a molar ratio of 1.4:1, VBS⅐vinculin. Lack of the full NMR resonance assignment of the very large vinculin protein precludes any residue by residue-based interpretation of the data, allowing us to assess only the global changes in the spectra. Nonetheless, superposition of the NMR data revealed significant shift changes and the appearance of numerous new cross-peaks in talin-VBS3-and ␣VBS-bound [ 1 H, 15 N]vinculin versus unbound [ 1 H, 15 N]vinculin (Fig. 3, A-C). This implies that vinculin undergoes significant changes in its overall conformation when bound to these VBSs. The extent of the changes upon binding was especially evident in the superpositions of ␣VBSversus talin-VBS3-bound full-length [ 1 H, 15 N]vinculin, which suggests that talin-VBS3 and ␣VBS also provoke at least some unique conformational changes in the overall struc-ture of vinculin (Fig. 3, D and E). Therefore, the VBSs of ␣-actinin or talin are sufficient to disrupt intramolecular interactions that hold vinculin in its inactive state, and appear to do so by provoking novel changes in the overall conformation of full-length vinculin.
The VBSs of Talin and ␣-Actinin Promote Binding of Vinculin to F-actin-When vinculin is activated it becomes competent to bind to F-actin, which requires severing of the head-tail interactions of vinculin (21,23). The ability of ␣VBS and talin-VBS3 to provoke significant changes in the structure of vinculin, and to disrupt the VH-Vt interactions, suggested that binding of these VBSs might also facilitate the binding of vinculin to F-actin. To test this hypothesis, ␣VBS and talin-VBS3 were incubated with full-length vinculin (at molar ratios ranging from 1:1 to 20:1, VBS:vinculin), and these complexes were then incubated with F-actin. As expected, full-length vinculin bound poorly to F-actin (Fig. 4). By contrast, a large proportion of ␣VBSor talin-VBS3bound vinculin was capable of binding to and co-sedimenting with F-actin, such that as much as 55% (␣VBS-bound, Fig. 4A) or 45% (talin-VBS3-bound, Fig. 4B) of vinculin became competent to bind to F-actin. Again, the increased F-actin binding potential of ␣VBS-bound vinculin correlates with its higher affinity for full-length vinculin versus talin-VBS3. Regardless, these findings establish that the VBSs of ␣-actinin and talin are indeed sufficient to activate the latent F-actin binding potential of vinculin, a hallmark of vinculin activation in adhesion complexes.
The VBSs of Talin and ␣-Actinin Selectively Compromise the Contacts of Vinculin at Sites of Focal Adhesions-The ability of the VBSs of talin and ␣-actinin to sever the head-tail interactions of vinculin suggested that these VBSs would also bind to vinculin in cells and would thus disrupt its ability to link to talin and ␣-actinin at sites of focal adhesions. To test this notion, Swiss-3T3 cells were microinjected with ϳ1,000 molecules of talin-VBS3 or ␣VBS peptides per cell and were then assessed for changes in the localization of endogenous vinculin, talin, and ␣-actinin, which concentrate at sites of focal adhesions, and for changes in the actin cytoskeleton by staining with phalloidin. As talin-VBS3 (C) peptides were incubated with fulllength vinculin at the indicated ratios of vinculin: VBS for 20 min at ambient temperature. Samples were then incubated for 1 h with F-actin as described previously (29), and samples were centrifuged into pellet (containing polymerized F-actin) and supernatant fractions, and equal volumes of these fractions were analyzed by SDS-PAGE. F-actin and vinculin were visualized by staining the gels with Coomassie Blue. As expected, native vinculin did not bind to F-actin, whereas ␣VBSand talin-VBS3-bound vinculin did bind to F-actin. B and D, the percent of vinculin co-sedimenting with F-actin (P) versus unbound vinculin (S), as determined by densitometry, is shown as function of the molar ratio of vinculin:VBS. expected, in control microinjected Swiss-3T3 cells there were pronounced actin stress fibers and intense staining of vinculin, talin, and ␣-actinin at sites of focal adhesions (Fig. 5). By contrast, in cells microinjected with talin-VBS3 or with ␣VBS peptides, vinculin, talin, and ␣-actinin were rapidly relocalized to the cytosol or were diffusely distributed to the margins of cell membranes and were no longer associated with focal adhesions (Fig. 5). Further, VBS-microinjected cells essentially lost nearly all of their focal adhesions, and there was also depolymerization and a collapse of the actin cytoskeleton (Fig. 5), consistent with the role of vinculin in stabilizing cell adhesion contacts (5,6). As a net result there were large reductions in the overall volume and size of VBS-microinjected cells (Fig. 5). Collectively, these data suggest that these VBS peptides can function as efficient sinks for vinculin in cells, disrupting the associations of vinculin with talin and ␣-actinin and then with the actin cytoskeleton.
The profound effects of the VBSs of talin and ␣-actinin on the actin cytoskeleton of Swiss-3T3 cells suggested that they should rapidly compromise cell-matrix contacts. As expected, video microscopy demonstrated that control fibroblasts lacked any noticeable changes in their shape or attachments following microinjection (Fig. 6, top panels, and supplemental data, movie 1). By contrast, there was a rapid loss of focal adhesions and a marked retraction and reduction in cell size and volume in ␣VBS-microinjected fibroblasts (Fig. 6, middle panels, and supplemental data, movies 2 and 3) and in talin-VBS3-microinjected cells (Fig.  6, bottom panels, and supplemental data, movies 4 and 5). Again, microinjection of as little as ϳ1,000 molecules of VBS peptides/cell was sufficient to induce collapse of the cytoskeleton, suggesting efficient disruption of nearly all focal contacts. With time these rather catastrophic events lead to the deaths of all talin-VBS3-or ␣VBS-microinjected cells, whereas control-microinjected cells remained viable and continued to proliferate (data not shown). Therefore, disruption of the contacts of vinculin and subsequent collapse of the actin cytoskeleton triggers fibroblast cell death.
Collectively, these findings suggested that ␣VBS and talin-VBS3 peptides target vinculin and disrupt its functions at sites of focal adhesions. However, to prove this was not because of off-target effects, we evaluated the consequences of microinjection of the VBSs of talin and ␣-actinin into isogenic vinculin Ϫ/Ϫ and vinculin ϩ/Ϫ cells (7). Vinculin-deficient cells have similar numbers of focal contacts but are less spread and less adhesive than their wild type counterparts (7). As expected, microinjection of talin-VBS3 or ␣VBS peptides provoked profound cell retraction, a reduction in cell size and volume, and loss of focal adhesions in vinculin ϩ/Ϫ cells (Fig. 7A, and supplemental data, movies 6 and 7), and ultimately these cells died (data not shown). By contrast, microinjection of talin-VBS3 or ␣VBS peptides failed to induce cell retraction or affect the size of vinculin Ϫ/Ϫ cells (Fig. 7B, and supplemental data, movies 9 and 10), and indeed these cells remained viable and continued to proliferate (data not shown). Therefore, the VBSs of talin and ␣-actinin selectively target vinculin functions in adhesion complexes. Fur-FIGURE 5. Microinjection of talin-VBS3 or ␣VBS peptides disrupts vinculin, talin, and ␣-actinin localization, the actin cytoskeleton, and focal adhesions. Swiss-3T3 cells were microinjected with ϳ1,000 molecules of ␣VBS or talin-VBS3 peptides. After 15 min cells were fixed and then stained with TRITC-phalloidin to detect actin filaments, or with antibodies that detect vinculin, talin, or ␣-actinin. In control microinjected Swiss-3T3 cells vinculin, talin, and ␣-actinin were all concentrated at the ends of actin filaments at sites of focal adhesions. By contrast, in Swiss-3T3 cells microinjected with talin-VBS3 or ␣VBS peptides there was a marked redistribution of vinculin, talin, and ␣-actinin, a reduced number of focal adhesions, and depolymerization of the actin cytoskeleton. Scale bar, 10 m. Also note the marked reduction in cell size and cell volume of talin-VBS3 or ␣VBS peptide microinjected cells. ther, their ability to disrupt focal adhesions and sever contacts with the actin cytoskeleton abolishes the tensive forces that anchor the cell to the extracellular matrix, which are required for cell survival.

DISCUSSION
Vinculin binding to talin or ␣-actinin provides essential links for adhesion receptors with the actin cytoskeleton and, as underscored by the studies presented here, vinculin plays essential roles in stabilizing adhesion complexes. The binding of PIP 2 was thought to activate vinculin to allow it to bind to its other partners, by altering the conformation of Vt and disrupting the head-tail interactions of vinculin (21,25,26). However, PIP 2 impairs the associations of vinculin with actin (29), and the structures of inactive vinculin revealed that the PIP 2 binding site is occluded in its inactive state (12,15). Furthermore, mutants of vinculin defective in PIP 2 binding have no defects in recruitment to focal adhesions (28). Thus, it follows that other regulators must activate vinculin, and the studies presented here suggest that the VBSs of talin and ␣-actinin fulfill this role.
Several lines of evidence support a direct role for talin and ␣-actinin in activating vinculin. First, both talin and ␣-actinin avidly interact with vinculin at sites of cell adhesions (49,51,52). Further, both talin and ␣-actinin harbor VBSs within their central rod domains that bind to full-length vinculin with affinities that are higher than (␣VBS, 2 nM; Ref. 40) or are comparable to (talin-VBS1 and -VBS3, ϳ70 nM; Fig. 1) those of the vinculin head-tail interaction (ϳ50 -90 nM; Refs. 17,18). In accord with these findings, the crystal structures of inactive human vinculin (12) and those of the Vh1⅐talin-VBS and Vh1⅐␣VBS complexes (24,35,36,40,41) revealed that the binding site in the Vh1 domain of vinculin that interacts with the VBSs talin and ␣-actinin is not "cryptic" nor occluded by the head-tail interaction; rather it is readily accessible to bind to these VBSs. Finally, these VBSs fulfill all the criteria one would expect for triggers that activate vinculin as they 1) efficiently displace the head-tail interactions of vinculin (Fig. 2), 2) induce significant alterations in the conformation of vinculin (Fig. 3 and Ref. 40), 3) promote vinculin binding to F-actin (Fig. 4), and 4) specifically target vinculin in cells and have profound effects on focal adhesions and the actin cytoskeleton (Figs. 5-7).
The binding affinities of talin and ␣-actinin for vinculin in their native, inactive states is low (15,18,20), and this has suggested a combinatorial model of vinculin activation, where two or more ligands are  (7) were microinjected with 1ϫ microinjection buffer alone (control), or with ϳ10,000 molecules of talin-VBS3 or ␣VBS peptide. Cells were then individually followed with time by time-lapse video microscopy. At the indicated intervals the position and shape of the microinjected cells was documented. Note the very rapid movement of vinculin ϩ/Ϫ cells following microinjection with ␣VBS peptide and an overall retraction in their cell size. Similar findings were evident following microinjection with talin-VBS3 peptide (data not shown). Again, the outline of injected cells prior to their injection is indicated at t 10 s (blue shade) and at t 10 min (red shade). Black dots at each time point mark the edges of the injected cells prior to injection. By contrast, microinjection of talin-VBS3 or ␣VBS peptide had no effect on vinculin Ϫ/Ϫ cells. Representative images are shown. Scale bars, 20 m.

FIGURE 8. Structural alterations as effectors in adhesion junction signaling.
A model is proposed where the formation of adherens junctions by cadherin receptors induces signals, for example mechanical stress, that elicits structural alterations in the ␣VBS ␣-helix in the rod of the ␣-actinin dimer. The hydrophobic face of ␣VBS is buried in the R4 spectrin repeat in ␣-actinin in its inactive state (42), but such a signal is suggested to provoke the ␣VBS helix to swing out and to undergo changes in its length, so it can then bind to the N-terminal helical bundle of the Vh1 domain of vinculin. Binding of ␣VBS alters the structure of the Vh1 domain (40), and this abolishes intramolecular interactions between the head and tail domains of vinculin (see Fig. 2). In so doing, this allows vinculin to open up to bind to its other partners, in particular to F-actin (see Fig. 4). By electron microscopy it has been shown that each protomer of the ␣-actinin dimer binds to one vinculin molecule (19). A similar scenario is envisioned to apply to focal adhesions, where the engagement of integrin receptors could provoke alterations in the structure of the talin rod that allow its ␣-helical VBSs to swing out to bind to the Vh1 domain and to activate vinculin (24,35,41,43).
needed to sever the head-tail interaction (15,18,53). Indeed, the hydrophobic faces of the VBSs that insert between the ␣1 and ␣2 helices of the Vh1 domain of vinculin are buried in the cores of the helical bundles present in the rod domains of talin and ␣-actinin (40,41,43). Thus, we predict that structural changes in these bundles occur following the interactions of talin or ␣-actinin with integrin receptors and that these alterations would include an unfolding of the bundles to release the VBSs to allow binding to vinculin (Fig. 8). Indeed, the ␣-helices within the three-helical bundles of spectrin repeats that are found in ␣-actinin are known to unfurl and form stable intermediates following their exposure to mechanical stress (44). We speculate that a similar scenario also applies to the helical bundles of the rod domain of talin when they are exposed to mechanical stress, which is sufficient to induce the formation of focal adhesions, and to recruit vinculin, within seconds of the applied stress (45,46). Such models would then allow for rapid, high affinity binding of these VBSs to vinculin, and their severing of the head-tail interactions of vinculin would then allow vinculin to bind to its other partners, in particular F-actin (Fig. 8). Further, such talin-vinculin and ␣-actinin-vinculin interactions appear required to stabilize focal adhesions, as displacement of their interactions by the free VBSs of talin or ␣-actinin leads to catastrophic effects on the actin cytoskeleton and rapid cell retraction (Figs. 5-7).
The notion that the VBSs of talin and ␣-actinin are sufficient to trigger vinculin activation appears, at first glance, to contradict the studies of others that have shown that the talin rod has a low affinity for fulllength vinculin (15) and that the Vh1⅐talin-rod complex is effectively displaced by the Vt domain (18). Our studies also challenge the notion that vinculin activation, and specifically its ability to bind to F-actin, requires two or more ligands to efficiently disrupt its head-tail interactions (15,53), which has been proposed to include both Vh1-Vt and Vt2-Vt interdomain contacts (18). However, these studies evaluated the interaction of the entire native talin rod with vinculin, with the bias that the structure of the rod of talin changes very little in its native versus vinculin-bound state. The crystal structures of the native helical bundles of talin (41,43) and of ␣-actinin (42), versus those of the VBS⅐vinculin complexes (24,35,36,40,41), have revealed that this is not the case, where the VBSs of talin and ␣-actinin must unravel from their buried locations to bind to vinculin. Given that at least talin avidly interacts with vinculin in cells (52), we therefore propose that structural alterations of the VBSs of talin and ␣-actinin are initial and necessary events that allow these ␣-helices to swing out to bind to and activate vinculin (Fig. 8). We recognize that the ability of the VBS peptides to efficiently disrupt the contacts of vinculin with talin and ␣-actinin in cells (Fig. 5) demonstrates that their affinity is higher than that of endogenous talinand ␣-actinin-VBS-vinculin interactions in cells. Nonetheless, the fact that this displacement occurs establishes central roles for these VBSs in mediating these interactions and in regulating actin dynamics in cells.
In cells, vinculin associates with actin at sites of focal adhesions or adherens junctions (54,55). In its native state, the affinity of vinculin for actin is rather low (ϳ1 M; Ref. 23), and others have suggested that talin-VBS3 (34) and pVR, a VBS of unknown significance (29), are not sufficient to activate the latent actin-binding potential of vinculin. However, these studies were performed by first incubating G-actin with vinculin and then testing the effects of talin-VBS3 or pVR, whereas ours tested the ability of talin-VBS3-or ␣VBS-bound vinculin to then bind to F-actin. We believe the latter schema more accurately recapitulates the chain of events in cells in vivo, and our studies clearly indicate that these VBSs can indeed activate the latent actin binding potential of vinculin. Evaluating the affinities of talin-vinculin-actin and ␣-actinin-vinculinactin interactions at sites of focal adhesions in cells, for example by fluorescence resonance energy transfer, should resolve the order and affinities of these interactions.